BATC Forum

These pallets are from recently de-commissioned transmitters that were built in 2017, so unlike a lot of surplus, they are pretty much state of the art.
The pallets use a single BLF888A with bias, input and output broadband matching originally specced for 470-850MHz. Not surprisingly for a broadband device, they work fine at 437MHz as-is and if you have the knowledge and test gear its quite possible you could improve the (already adequate) gain and output power at 432 or 437MHz.
All I have done is mount a pallet, feed it DC (50V at up to 15A, bias (around 3.5-3.7Vgs for Ids 1.3A INCREASE Vgs ONLY WHILE MONITORING CURRENT DRAWN as the onset is very sudden. It draws about 10mA so use a robust source) and RF 1-2 watt and measured the significant output.
115W of DVB-T at impressive linearity - 150W if pushed a bit harder.
220W DBVS2 at acceptable linearity.
At least 350+ W FM/CW/SSB two-tone. (Data sheet shows 400W)
Data sheet here https://www.ampleon.com/products/uhf-br ... F888A.html well worth a read.
The board is a mere 142x60mm and about 15mm high.
A copper heat-spreader would be a good idea if you expect to run full power, (the FET base protruded about 1.2mm below the PCB base), there is also a wire currently underside that feeds VD from the single stand-off feed-point to the other side of the FET) and a BIG heatsink with a LOT of fan(s). The Vgs may also need some tiny clearance under the board as its flush with the ground-plane underneath.
For the inveterate experiments amongst you, it crossed my mind that as the device is rated 300MHz-1GHz (and is so a broadband device rather then the frequency matched stuff we sometimes see in UMTS pallets) it might well work rather well at 23cms. W6PQL has a PCB for a very similar device, itself rated at 900MHz-1.3GHz https://www.nxp.com/docs/en/data-sheet/MRF13750H.pdf. Its on my list to try one and see....

£40 to BATC members only. Plus delivery £5 UK only, for anywhere else please enquire about costs first. Payment by Paypal F&F or bank transfer.
Any questions, please ask here...but message me if you want to buy one.
Gareth

The idea behind these amplifiers is to use them for DVB-T on 70cm so I thought I would do some tests on the driver unit. I can get 300W CW out of mine before chickening out. More power can be had, but overdrive can result in unwanted smoke emission.

The results below are for the driver pallets which use a BLF888D, a Doherty device to drive 6 final pallets. That’s the one I have shown on the net with the weird output matching lines. The results should not be that different from the non-Doherty pallets. I did test those too. See later.

Running these pallets at 300W for a TV QSO isn’t likely to be a good idea and the spectrum spreading matters. What works for CW and even SSB isn’t good enough for OFDM or QPSK.

• Headline 75W to 100W depending on what’s considered acceptable an level of spectral spreading.

The tests

I used ta very linear driver PA driven from a Lime Mini and PD4. The input signal needs to be better than the output for results to be useful. In this case the input had better than -50 dB shoulders.

The drain voltage was 50V. In testing I learned why we must bias both devices. I biased at 650mA per side. Turning either peak or main side off degraded the spectrum considerably. At 50W and below it is as good as the drive signal.

DVB-T

I tested with DVB-T as it is a bit harder than DVB-S and most likely mode for terrestrial use. At 75 W where the drive signal was still very clean, the PA was just starting to show shoulders – 45dB, the current was 8A.
Turning off bias to the main side of the BLF888D reduced the current to 4A but had almost no impact on the power output. Turning off peak side bias (while retaining main) greatly reduced the output. Turning off both reduced the output and looked awful, so it is probably obvious but confirms we need both peak and main biased to get a good spectrum for DVB-T but if we want just efficiency for CW we probably only want to use peak, or at least different biasing. I did a quick test with increased biasing and it didn’t seem to help very much. Reducing the peak bias current a bit didn’t make it much worse at the 100W level, nor did increasing it. The main side bias was a bit more sensitive but it seems as long as there is some current and it is in AB1 it is OK.
The spectrum got rapidly worse above 75W. By 100W, which requires 10A of drain current, the shoulders have degraded to -30 dB. That might be improved a little with optimised biasing but the device spec is only -39 dB at this level and the peak to mean won’t be the same for 333kHz vs 6 MHz so I am not sure we can make a direct comparison. At 333kHz or even 500 kHz the 100W signal is still going to be narrower than even an ultra clean 1 Ms transmission.
At 180W it is getting much worse, shoulders – 20 dB.
I tested up to 600 MHz and it is very consistent so there is probably not much benefit to be had from tuning.

DVB-S to follow:

DVB-S2
I hadn’t realised quite how much higher the lime output is with SVB-S2 compared to DVB-T so first test was 200W and 14A. Shoulders -35 dB.
A more sensible output, 110W and 10.4A shoulders – 40 dB.
And 75W again, which needed 8A. I had to adjust the analyser attenuation to see the shoulders which are -50dB. This needs 2W input.
Mike

And finally, here is a test of the BLF888A.

This is 1 MHz of DVB-T 64 QAM at 200W RF and about 500W DC input. There is 70 dB of attenuation in line and the resolution bandwidth 10 kHz, video bandwidth 3kHz – PSU 50V at 15A

Some more plots at lower output, 50W and 100W:
Mike

Thanks Mike for posting those useful images. I was at the Martlesham Microwave Round-table yesterday and took the opportunity to run the amp up. We reached 300W CW with negligible evidence of gain compression. Pretty much as per the data sheets suggest.
7 now requested, still a few left.
Gareth

As of NOW, Tuesday 7:20 p.m., no further pallets are available, subject to confirmation of payment (or not) there may be two more. Will post here if they become available. I do know of 'more in the system', so do not despair just yet...
Gareth